You are on page 1of 7

INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 6, ISSUE 04, APRIL 2017 ISSN 2277-8616

Occupancy Pattern Of A Forest Dependent Bird


Among Coastal Forest Fragments In Northeast
Tanzania
Robert B. Modest

Abstract: The loss of biological resources in the coastal forests of eastern Tanzania is alarming. This is due to human related activities such as
vegetation clearing for agriculture and intensive livestock grazing. By their nature, these activities affect forest dependent birds through destroying
habitat and or blocking migratory corridors, and thus interrupting site occupancy pattern. The aim of this study therefore, was to determine whether
habitat degradation along the Tanzanias north eastern coast affects site occupancy patterns of forest dependent birds among forest fragments and the
associated savannahs. Lowland Tiny Greenbul, a forest dependent bird was used as a model. The data was collected along transects set inside the
forest fragments and along the neighboring matrices. The collected data was then used to build site occupancy probability models using the software
Presence. The results revealed that ideal undisturbed habitat positively influenced both the relative abundance and site occupancy probability of the
model bird indicating the significance of maintaining habitat in their natural state for the welfare of forest dependent species and the broader
biodiversity. This study emphasizes minimizing human pressures in the forests and the matrices for the persistence of the species.

Key Words: Birds, Forest Clearing, Biodiversity, Savannahs, Migratory Corridors.


1 INTRODUCTION There are four vegetation complexes in the area; "1) A


The Tanzanias coastal forests which are part of the coastal heterogeneous forest-savannah-grassland mosaic, 2) the
forests of eastern Africa worlds biodiversity hotspot are coastal forests, 3) a shoreline with salt flats, coastal fringe
globally renowned for their high level of biodiversity and forests, herbaceous dune vegetation and mangrove forests,
endemism in both flora and fauna [1], [2], [3], [4]. The and 4) a maritime ecosystem". Some common tree species
biodiversity in these forests however have traditionally faced include Adansonia digitata, Afzelia quanzensis, Pteleopsis
severe anthropogenic pressures emanating from activities myrtifolia, and Synaptolepis kirkii [1]. Generally, the vegetation
such as pole collection, clearing for agriculture and tree felling along the landscape show different levels of degradation
for timber [2], [5], Fig 1. These activities which are deleterious following persistence human interference with loss of forest
in nature have been reported to pose enormous negative cover in protected areas being lower than those in unprotected
effects on avian biodiversity elsewhere, such as reducing segments [11]. Moreover, the human induced pressure
species richness and abundance [6], [7], as well as intensifies in a south-north orientation, whereby the northern
interrupting species site occupancy pattern [4], [8]. For part is heavily degraded due to intensive livestock grazing,
fragmented habitat such as those along Tanzanias coastal human settlements, tree cutting, and cultivation of semi-annual
forests on the other hand [2], the possibility of these activities and annual crops. As these activities can be detrimental to
driving contiguous species populations into island sub- some groups of birds, investigating occupancy pattern of forest
populations is high [9]. Determining occupancy pattern of dependent species is essential to understand the effects of
wildlife in habitat that have suffered fragmentation such as human induced pressures on their populations. The aim of this
those in the coastal forests of eastern Tanzania is thus study thus, was to find out whether site occupancy patterns of
essential for understanding changes in their status over time. forest dependent birds differs among habitat depicting different
The findings can provide useful information in decision making degree of degradation along the aforementioned landscape.
and on setting priorities for species management [10]. The study used the Lowland Tiny Greenbul (Phyllastrephus
Therefore, this study was designed to investigate the site debilis) as a model. This bird is a forest specialist species
occupancy pattern of a forest dependent bird species among which is mainly restricted to well-developed forested habitat,
fragmented coastal forests in north eastern Tanzania. The and preferring forest interior [12]. The species range covers
study covered the landscape stretching between Pangani and the moist lowland forests in eastern Africa, but also may occur
Wami Rivers along the Tanzanias north east coast. Along this in dense bushes around forest edges, and its main diet
landscape are scattered and fragmented coastal forests comprises butterflies, bees, wasps, locusts and ants [13]. As it
mostly deteriorated, but with some remnants of healthy and is well established that this model bird prefer ideal intact
vibrant patches [2]. habitat, information on how the species is affected by severity
of habitat degradation however, is not available. The
hypothesis tested stated that, levels of habitat degradation
among coastal forests along the north eastern Tanzanias
coast would have no influence on site occupancy pattern of
the model bird under investigation.

_________________________

Robert B. Modest, Department of Wildlife management,


Sokoine University of Agriculture, Tanzania, Email:
robertbmodest@yahoo.com
67
IJSTR2017
www.ijstr.org
INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 6, ISSUE 04, APRIL 2017 ISSN 2277-8616

distances of 150 m from each other. In Zaraninge which is


the biggest of the study forests there were nine transects,
whereas the other forests had a fewer number of transects as
follows; Msubugwe (six), Gendagenda (five), and Kwamsisi
(four). Moreover, a single transect was placed in the matrices
separating Zaraninge and Kwamsisi, Kwamsisi and
Gendagenda, and Gendagenda and Msubugwe forests
respectively. The transect length within the forests ranged from
100 m to 500 m, while those in the matrices were 4000 m
each. Data gathering was done into two blocks of time from
07.00 hours to 10.30 hours for the morning session, and from
03.30 hours to 6.30 hours for the evening session. Moreover,
data collection involved quantification of levels of habitat
disturbance along each transect in order to describe the
characteristic of each study site termed as site-specific
covariates [15], [16]. For the case of forest fragments, coding
of levels of habitat disturbance based on counting number of
freshly cut tree stumps along transects (which were then
summed-up at the end of the study). The coding followed the
scale as follows: No stump, = non disturbed forest; 1 to 10
Fig.1: Depiction of the prevailing situation in the study area stumps = lightly disturbed forest; 11 to 20 stumps = disturbed
forest; and more than 20 stumps = highly disturbed forest. To
2 METHODOLOGY ensure that the stumps were only counted once during the
course of the study, the wall paint was used to mark every new
2.1 Study Area stump during site re-visit. On the other hand, coding levels of
The study was carried out within the East Africa Coastal habitat disturbance in the savannah matrices based on noting
Forests worlds biodiversity hotspot in North Eastern Tanzania. number of livestock and/or human trails along transects as
The geographic coordinates are 616'42.94 to 616'57.65S, follows: No trail = undisturbed Savannah; 1 to 4 trails = lightly
and 3832'08.35 to 3851'17.37E [2], [4] Fig. 2. The area disturbed savannah; and 5 or more trails = highly disturbed
receives a high peak of rainfall from March to May, and there savannah. For the case of this study; to distinguish human
is a shorter spell of rainfall from October to December [1]. trails from trails created by wildlife a human trail was defined
Further description of the study site is provided in [4]. as the one with some signs of tree cutting and leading to a
water source, to a nearby village, or to pitsaw camps.
Moreover, to distinguish livestock trails from those created by
wildlife, each trail was scanned for presence of cattle dungs. In
addition to these criteria, the researchers experience of the
study area, and a review of previous publications such as that
by Burgess & Clarke [17], as well as Hassan et al. [18] were
useful in the characterization of the sites. The site
characterization described above resulted into seven site-
specific covariates as shown under Table 1. These covariates
are the ones that were used in modeling the birds site
occupancy probability along with the abundance, and
seasonality data.

Table 1: Study sites characterization into site-specific


covariates based on their levels of disturbance.
Fig. 2: Map of the study area. Dark grey-shaded areas
Abbreviation for
indicate study forests. Dashed lines show transects in the Study site
Site-specific covariate
site-specific
matrices. name
covariate
Zaraninge Non Disturbed Forest NDF
2.2 Data Collection Kwamsisi Disturbed Forest DF
Data collection commenced between October 2010 and June Gendagenda Lightly Disturbed Forest LDF
Msubugwe Highly Disturbed Forest HDF
2013. The data was collected from four coastal forests namely
Zaraninge/Kwamsisi
Zaraninge (42.7 km2, plateau only), Kwamsisi (4.06 km2), matrix
Undisturbed Savannah UNS
Gendagenda (10.97 km2), and Msubugwe (22.32 km2), Fig. 2. Kwamsisi/Gendagend Lightly Disturbed
LDS
The data was collected monthly, whereby an observer a matrix Savannah
recorded number of individuals of the study species along Gendagenda/Msubug Highly Disturbed
HDS
transects established within the abovementioned forests. To we matrix Savannah
ensure that the species habitat is covered adequately,
transects were located in the interior habitat as well as within
10 m from forest edges [14]. In order to minimize the chances
of double counting the birds, transects were placed at inter-
68
IJSTR2017
www.ijstr.org
INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 6, ISSUE 04, APRIL 2017 ISSN 2277-8616

2.3 Data analysis Table 2: Relative abundance of the species per site: The
The bird survey data was used to obtain the relative relative abundance is regarded as the actual number of
abundance of the species in each site regarded as total individuals recorded per site.
number of individuals per site [19]. On the other hand, the
same bird survey data was coded into presence/absence for Relative
Site Size (km2)
modelling the site occupancy probabilities using the software Abundance
Presence [15]. In order to meet the closure assumptions Zaraninge (Plateau only) 42.7 415
required for modelling using the software Presence, the Msubugwe 22.32 148
presence/absence data was sorted into seasons and only the Gendagenda 10.97 97
data collected during the middle of each seasons was Kwamsisi 4.06 64
analysed [15], [20]. The study area experiences four seasons, Zaraninge/Kwamsisi matrix - 22
for this matter, analysis focused on data collected during
Kwamsisi/Gendagenda matrix - 5
November to December (middle of the short rainy season),
April to May (middle of the long rainy season), February to Gendagenda/Msubugwe matrix - 0
early March (middle of the short dry season), and July to
August (middle of the long dry season). These survey seasons Table 3 presents the a priori models used in modelling site
were treated as survey specific covariates during modelling specific and survey specific covariates against the survey
[20]. Then, using the site specific covariates defined under data. Based on prior knowledge and literature review, nine
Table 1, and the survey specific covariates defined as models were used to determine the influence of the covariates
p(survey), nine a priori models were established. For each on the study species.
season three models were evaluated: 1) a constant model
which treated both occupancy and detection probabilities as Table 3: The a priori models used on determining occupancy
being constant; 2) models where occupancy was a function of (), and detection probabilities (p) of the study species. Nor.
site-specific covariates while detection probability was a par = number of parameters. See Table 1 for description on
function of survey-specific covariates; and 3) models where abbreviations of the model covariates.
both occupancy and detection probabilities were functions of
site-specific covariates, Table 3. While the site-specific Model No.par. Description
covariates used are those defined under Table 1 above, the Both occupancy and detection
(.),p(.) 2
probability are constant.
survey-specific covariates were obtained based on the Occupancy influenced by non-
seasons covered during bird survey, see for example disturbed forest, detection
(NDF),p(Survey) 4
Mackenzie et al. [20] for details. On defining the nine a priori probability influenced by survey
models in Table 3, only those covariates that were believed season.
could potentially explain occupancy probability () and Occupancy influenced by lightly
disturbed forest, detection
detection probability (p) of the study species were evaluated (LDF),p(Survey) 4
probability influenced by survey
[20, 21]. The software Presence ver. 6.1 was used to build the season.
site occupancy probability (), and detection probability (p), Occupancy influenced by
following the single-season single-species approach through (DF),p(Survey) 4
disturbed forest, detection
the application of the likelihood theory [15], [20]. Model probability influenced by survey
selection to evaluate support of the data, and to assess the season.
Occupancy influenced by heavily
strength of each site and survey specific covariates on disturbed forest, detection
influencing the species based on Akaike Information Criterion. (HDF),p(Survey) 4
probability influenced by survey
Following this selection procedure, models with delta Akaike of season.
2 were considered to have support in the data and were (NDF),p(NDF) 5
Both occupancy and detection
treated further on averaging. Averaging of a competing set of influenced by non-disturbed forest.
Both occupancy and detection
candidate models was achieved using the following formula: (LDF),p(LDF) 5 influenced by lightly disturbed
forest.
Both occupancy and detection
(DF),p(DF) 5
influenced by disturbed forest.
Both occupancy and detection
Where = Akaike weight for model j, = overall detectability (HDF),p(HDF) 5 influenced by heavily disturbed
forest.
or occupancy for model j, = overall detectability or
occupancy for the averaged model [22].
Models were assumed to have support in the data if their delta
Akaike were 2. Based on this criterion, seven models were
3 RESULTS AND DISCUSSION supported for the long dry season, whereby, non-disturbed
forest (NDF) and season effects p(Survey) were the most
3.1 Results influential with the Akaike weight of 0.19. For the short dry
A total of 751 individual birds were recorded. Table 2 presents season, only the constant model was supported. Three models
the relative abundance of the species in each study site. were supported during the long rainy season, and a constant
model had the highest weight. For the short rainy season, on
the other hand, four models were supported, whereby, non-
disturbed forest (NDF), and p(survey) covariates were the
most influential accounting for the highest Akaike weight of
0.31 (Table 4). The overall averaged site occupancy
69
IJSTR2017
www.ijstr.org
INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 6, ISSUE 04, APRIL 2017 ISSN 2277-8616

probabilities ( ), for the modelled seasons were; Long dry congestion in this forest. Msubugwe forest, in addition to tree
season (0.38), Short dry season (0.35), Short rainy season cutting pressure in its core habitat [12], [18], faces severe
(0.60), and Long rainy season (0.53). pressure from grazing livestock in the surrounding matrices
(personal observation). The pressure in the matrices probably
Table 4: Model selection inference for the Tiny Greenbul is restraining individuals inside the forest, and the forest is
occupancy () and detection probability (p) based on certainly acting like a remote island for the species [23].
presence/absence survey of the coastal forests in north east Hence, the higher relative abundance of the Lowland Tiny
Tanzania. The best model in each season is shown at the top Greenbul (a forest specialist that prefer natural habitat) in
of the lists. AIC = difference between maximum and Msubugwe forest might not imply association with quality
minimum Akaike values of competing models; W = Akaike habitat; rather individuals are unable to move outside due to
weight. the high level of habitat disturbance in the surrounding
matrices c.f. [24]. The more important observation is that,
Model / Season AIC w about three quarter of individual birds were recorded in the
south eastern side of Msubugwe. The south eastern part of
Long dry season
this forest is relatively less disturbed compared to the western
(NDF),p(Survey) 0 0.19 part. Therefore, it seems that the birds are avoiding the
(NDF),p(NDF) 0.18 0.17 sections of the forest that are hit hard by tree cutting and
(DF),p(Survey) 0.99 0.12 vegetation clearing (see Fig. 3 for impression of the habitat
(LDF),p(Survey) 0.99 0.12 degradation in the study area). Maclean et al. [25] also found a
similar phenomenon on water dependent birds in one of the
(HDF),p(Survey) 0.99 0.12
wetland in Uganda. This study by [25] reported higher
(LDF),p(LDF) 1.78 0.08 concentration of water dependent birds in a number of small
(HDF),p(HDF) 1.78 0.08 patchy swamps, and they attributed this to lack of alternative
Short dry season habitats as the area was facing high anthropogenic pressure
(.),p(.) 0 0.42
and habitat fragmentation. Newmark et al. [26] also observed
a similar phenomenon in West Usambara Mountain north east
Short rainy season
Tanzania, where they reported three forest specialist bird
(.),p(.) 0 0.31 species failing to cross a cleared forest gap of just less than
(NDF),p(survey) 1.03 0.18 15 meters. Therefore, the Ugandas wetland, and the West
(NDF),p(NDF) 1.58 0.14 Usambara Mountain scenarios might also apply to the
Msubugwe case. Thus, it can generally be proclaim that,
Long rainy season
Msubugwe forest perhaps is currently in a stated of
(NDF),p(survey) 0 0.31 inhospitable remote island for the Lowland Tiny Greenbul,
(NDF),p(NDF) 1.13 0.17 and that individuals are unable to disperse as the forest is
(HDF),p(HDF) 1.13 0.17 surrounded by highly degraded matrices intolerable by the
(LDF),p(Survey) 1.35 0.16
species [23],[27].

3.2 Discussion
The higher relative abundance of the Lowland Tiny Greenbul
was recorded in Zaraninge, a forest which was categorized as
non-disturbed. However, this forest is not 100% undisturbed
as it had suffered severe tree cutting in previous years. The
forest was put under a non-disturbed category because during
data collection there was no freshly cut trees (c.f. plateau
segment of the forest). Additionally, all logs and stumps from
the previous logging shocks had decomposed, thus, the forest
looked almost natural. Zaraninge forest used to be a forest
reserve, under which people could easily accesses its
resources owing to low protection status. But, currently the
plateau section of this forest is under Saadani National Park,
and thus, is managed under strict rules with no extractive use
allowed therefore, the recovery of this segment of the forest
can be attributed to this fact. Surprisingly, Msubugwe which
was categorized as a highly degraded forest ranked the
second in terms of relative abundance of the species. This Fig. 3: An area of a previously closed section of the forest that
brings in an ambiguity. As a matter of fact, the relative was converted into a pineapple field. Some native trees that
abundance of the model bird (a forest specialist) in Msubugwe were left behind after clearing can still be seen at the
forest was supposed to be even lower than that recorded in background, but due to change in habitat type, they eventually
Gendagenda (though smaller in size) taking into account die.
that Gendagenda is a less disturbed forest compared to the
former. The fact that Msubugwe is surrounded by highly For the occupancy probability analyses, the covariate non-
degraded savannahs, the higher relative abundance of the disturbed forest seemed to be more important for the species.
bird (a forest specialist) could be a result of individual This covariate appeared twice in the best models across the
70
IJSTR2017
www.ijstr.org
INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 6, ISSUE 04, APRIL 2017 ISSN 2277-8616

four seasons modeled. In addition, all other models that seasons, or the data collected on a single season survey
contained non-disturbed forest as a site-specific covariate bases should not be extrapolated across other seasons.
always had higher Akaike weights (Table 4), and this agrees
with the abundance data discussed above. These phenomena 4. CONCLUSION
therefore, are an indication of the significance of natural In conclusion, both the abundance and modelling data suggest
habitat for the study species Fig. 4. The phenomena also calls a susceptibility of the model bird to habitat degradation. This
for a need of maintaining the forests in their primary follows, as it was observed that, ideal non-disturbed habitats
undisturbed condition for the welfare of the species as favourably influence the species relative abundance and site
such, the species is likely to disappear if habitat loss in the occupancy probability. Moreover, this study is concerned with
study system continues, cf. [28]. Species with specialized the welfare of the model bird in Msubugwe, as the species is
habitat requirements such as the Lowland Tiny Greenbul are probably captured in this highly degraded forest and
sensitive to habitat disturbance cf. [29], thus, the association of individuals are unable to move out following degradation of
the Lowland Tiny Greenbul with undisturbed habitat within the matrices connecting this forest to other better sites. Thus,
coastal forests of northeastern Tanzania rings an alarm should habitat degradation continues, there is a possibility of
regarding the persistence of the species in the area. This driving the species to local extinction as there is similar cases
follows as it is reported that, habitat degradation within Vikindu already reported in other region of the coastal forests of
forest in the southern neighborhood of Zaraninge caused local Tanzania. On the other hand, this study have revealed that the
extinction of the Sokoke pipit [30], a bird species whose model bird is possibly not sedentary as previously thought, as
general habitat specialization is comparable to the Lowland its site occupancy probability seemed to vary by seasons as
Tiny Greenbul. such, Lowland Tiny Greenbul is perhaps able to undergo local
migration along the landscape via well-established savannahs.
Since this study focused only on bird survey, a follow-up study
using advanced technologies such as telemetry is recommend
to establish the species dispersal pattern in the area. For
proper management of the species within the coastal forests in
eastern Tanzania, habitat restoration in the Msubugwe forest is
recommended to enhance individual occupancy of all the
niches available. Lastly, the on-going tree cutting within the
forests and matrices should be stopped through applying strict
control measures for the benefit of the study species and the
general biodiversity.

5 ACKNOWLEDGMENTS
The funds to carry out this study were provided by SUA-VLIR
Programme. Tanzania National Parks offered free entry
permission to collect data inside Saadani National Park. The
Fig. 4: Lowland Tiny Greenbul Phyllastrephus debilis, a victim field assistants are thanked for their support during data
of habitat degradation in the coastal forests of north eastern collection.
Tanzania.

Moreover, the occupancy probability of the model bird under 6 REFERENCES


the current study seemed to be influenced by seasonality. This [1] U. Bloesch and F. Kltzli, The vegetation of the
is because the highest overall averaged occupancy probability Saadani National Park and possible conservation and
was attained during the short rainy season, whereas, the management strategies, unpublished report, Wildlife
lowest was during the short dry season. This tendency i.e. the Division, Dar es Salaam, Tanzania, 2002.
discrepancy in the site occupancy probability across the
seasons could be implicated to the breeding circle of the [2] E.T. Azeria, I. Sanmartin, S. As, A. Carlson and N.
species which is documented to begin during October [13]. Burgess, Biogeographic patterns of the East African
The bird activity and singing during this season probably coastal forest vertebrate fauna, Biodiversity
increases and birds are presumably displaying for finding Conservation, vol.16, pp. 883912, 2007.
mates where it is easy to detect. On the other hand, the
discrepancy could also be an indication that the Lowland Tiny [3] A.A. Rija, A. S. Mgelwa, R.B. Modest and S.N.
Greenbul is probably not sedentary as previously reported, for Hassan, Composition and Functional Diversity in Bird
example, by BirdLife International [30]. The species is perhaps Communities in a Protected Humid Coastal
able to undergo local migration by swapping habitat among Savannah, Advances in Zoology, vol. 2015, Article ID
forests and matrices though majority of individuals remain 864219, 7 pages, 2015.
within forested habitat. The significance of these findings is
that, results derived from a single season survey of the [4] R.B. Modest, S.N. Hassan, and A.A. Rija, Spatial
species in anyone locality within the coastal forests of east metrics effect of forest fragmentation on forest bird
Africa cannot be generalized to explain the status of the abundance and site occupancy probability: the
species in the area. Thus, any future survey in the area influence of patch size and isolation, Ostrich,
intending in estimating the abundance of the species for http://dx.doi.org/10.2989/00306525, 8 pages, 2016.
example, should be designed as a repeated survey across
[5] A. Ahrends, N.D. Burgess, S.A.H. Milledge, M.T.
71
IJSTR2017
www.ijstr.org
INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 6, ISSUE 04, APRIL 2017 ISSN 2277-8616

Bulling, B. Fisher, J.C.R. Smart, G.P. Clarke, B.E. studies with imperfect detection: examples and
Mhoro, and S.L. Lewis, Predictable waves of software, Ecological Applications, vol. 17, no. 1, pp.
sequential forest degradation and biodiversity loss 281290, 2007.
spreading from an African city, Proceedings of the
National Academy of Sciences of the United States of [17] N.D. Burgess and G.P. Clarke, Eds. Coastal Forests
America, vol.107, pp. 1455614561, 2010. of Eastern Africa. IUCN Forest Conservation
Programme, Gland, Switzerland and Cambridge,
[6] R. Duraes, L. Carrasco, T.B. Smith and J. Karubian, England, 2000.
Effects of forest disturbance and habitat loss on
avian communities in a Neotropical biodiversity [18] S.N. Hassan, A.R. Salum, A.A. Rija, R.B. Modest, J.R.
hotspot, Biological Conservation, vol. 166, pp. 203 Kideghesho, and P.F. Malata, Human-induced
211, 2013. Disturbances Influence on Bird Communities of
Coastal Forests in Eastern Tanzania, British Journal
[7] N.E. Otieno, N. Sajita and D. Shitandayi, Response of Applied Science and Technology, vol. 3, no.1, pp.
of a globally endangered canopy insectivore to habitat 48-64, 2013.
degradation in an East African tropical rainforest: The
role of differential forest protection levels, [19] S.L. Winter, S.D. Fuhlendorf, B.L. Smith and W.
International Journal of Biodiversity and Conservation, Street, Breeding Bird Communities Associated With
vol. 6, no.3, pp. 290-300, 2014. Tallgrass Prairies in Southeast Nebraska,
Transactions of the Nebraska Academy of Sciences
[8] A.B. Franklin, B.R. Noon and T. Luke, What Is and Affiliated Societies, vol. 33, pp. 3945, 2013.
Habitat Fragmentation?, Studies in Avian Biology,
vol. 25, pp. 2029, 2002. [20] D.I. Mackenzie, J.D. Nichols, A.J. Royle, K.H. Pollock,
L.L. Bailey and J.E. Hines, Occupancy estimation and
[9] R. Frankham, Inbreeding and Extinction: Island modeling: inferring patterns and dynamics of species
Populations, Conservation Biology, vol. 12, no.3, pp. occurrence. Elsevier Publishing, 2006.
665675, 1998.
[21] W.D. Newmark and T.R. Stanley, Habitat
[10] J.R. Sauer, U. Patuxent and B.F. Road, Using Multi- fragmentation reduces nest survival in an Afrotropical
Species Occupancy Models in Structured Decision bird community in a biodiversity hotspot, Proceedings
Making on Managed Lands, The Journal of Wildlife of the National Academy of Sciences, vol. 108, pp.
Management, vol. 77, no.1, pp. 117127, 2013. 11488-11493, 2011.

[11] K. Tabor, N.D. Burgess, B.P. Mbilinyi, J.J. Kashaigili [22] K.P. Burnham and D.R. Anderson, Model Selection
and M.K. Steininger, Forest and woodland cover and and Multinomial Inference: A Practical Information-
change in coastal Tanzania and Kenya, circa 1990 to Theoretic Approach (second Ed), Springer, New York,
circa 2000, The Journal of East Africa Natural 2002
History, vol. 99, no. 1, pp. 1945, 2010.
[23] R.H. MacArthur and E.O. Wilson. The Theory of
[12] R.B. Modest and S.N. Hassan, Species Composition Island Biogeography. Princeton University Press,
of Tropical Understory Birds in Threatened East 1967.
African Coastal Forests Based on Capture
Data, International Journal of Zoology, vol. 2016, [24] K.E. Sieving, M.F. Willson and T.L. De Santo, Habitat
Article ID 1390364, 9 pages, 2016. Barriers to Movement of Understory Birds in
Fragmented South-Temperate Rainforest, The Auk,
[13] L. Fishpool and J. Tobias, Tiny Greenbul vol. 113, no. 4, pp. 944-949, 1996.
(Phyllastrephus debilis, in Handbook of the Birds of
the World Alive, J. del Hoyo, A. Elliott, J. Sargatal, [25] I.M.D. Maclean, R.J. Wilson and M. Hassall,
D.A. Christie and E. de Juana, Eds, Lynx Editions, Predicting changes in the abundance of African
Barcelona, http://www.hbw.com/node/58002, 2016. wetland birds by incorporating abundanceoccupancy
relationships into habitat association models,
[14] J.C. Bibby, N.D. Burgess, A.D. Hill and S.H. Mustoe, Diversity and Distributions, vol. 17, pp. 480-490, 2011.
Bird Census Techniques, Academic Press, London,
UK, 2000. [26] W.D. Newmark, V.J. Mkongewa and A.D. Sobek.
Ranging behavior and habitat selection of terrestrial
[15] S.L. Rogers, J.A. Collazo and C.A. Drew, King Rail insectivorous birds in northeast Tanzania: implications
(Rallus elegans) Occupancy and Abundance in Fire for corridor design in the Eastern Arc
Managed Coastal Marshes in North Carolina and Mountains, Animal Conservation, vol. 13, pp. 474-
Virginia, Waterbirds, vol. 36, no.2, pp. 179188, 482, 2010.
2013.
[27] W.D. Newmark, The role and design of wildlife
[16] L.B. Bailey, J.E. Hines, J.D. Nichols and D.I. corridors with examples from Tanzania, Ambio,
Mackenzie, Sampling design trade-offs in occupancy vol.12, pp. 500504, 1993.

72
IJSTR2017
www.ijstr.org
INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 6, ISSUE 04, APRIL 2017 ISSN 2277-8616

[28] L. Bennun, C. Dranzoa and D. Pomeroy, The Forest


Birds of Kenya and Uganda, Journal of East African
Natural History, vol. 85, no. 1, pp. 2348, 1996.

[29] N.E. Otieno, N. Sajita and D. Shitandayi, Response


of a globally endangered canopy insectivore to habitat
degradation in an East African tropical rainforest: The
role of differential forest protection levels,
International Journal of Biodiversity and Conservation,
vol. 6, no. 3, pp. 290-300, 2014.

[30] Birdlife International, Species factsheet:


Phyllastrephus debilis. http://www.birdlife.org, 2014.

73
IJSTR2017
www.ijstr.org